JPH01262428A - Spectral measuring instrument - Google Patents

Abstract

PURPOSE:To eliminate the waste of a memory means and to facilitate the selective setting of effective picture elements and ineffective picture elements by taking only the necessary picture element output into a data processor by the action of a dip switch. CONSTITUTION:Plural pieces of spectral filters 1, 2 are discretely disposed on a solid-state image pickup element 3 of the instrument. The positions of the ineffective picture elements of the element 3 to the incident light transmitted between the filters 1 and 2 or the effective picture elements of the element 3 to the incident light transmitted through the filters 1, 2 are stored by the dip switch. Only the effective picture element output is taken into a microcomputer. The taking of the unnecessary picture element into the microcomputer is obviated at the time of taking the picture element output into the microcomputer by such constitution and, therefore, only the data on the effective picture elements is stored in the RAM of the microcomputer. The easy selective switching by the dip switch is enabled even if the positions of the effective picture elements and the ineffective picture elements change.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a measuring device for electrically measuring the spectral distribution of light to be measured by combining a plurality of spectral filters and an integral solid-state image sensor.

[Conventional technology]

As a method or apparatus for performing this type of spectroscopic measurement, a method using a spectral filter and a self-scanning solid-state image sensor such as a CCD image sensor is known, as disclosed in Japanese Patent Laid-Open No. 57-69222.

In addition, although the output of a solid-state image sensor (hereinafter referred to as CCD as an example) is normally taken into the data processing control circuit via an A/D converter, it is effective as a countermeasure when unnecessary pixels are included in the pixels of the CCD. Captures only pixel output,
In other words, in order to prevent unnecessary pixel outputs from being taken in, we have previously proposed a technique in which the conversion start pulse to the A/D converter is masked using a logic circuit.

[Problem to be solved by the invention]

In a solid-state imaging device for spectroscopic measurement in which such a spectral filter and a solid-state imaging device are combined, it may be difficult to cover the entire wavelength range to be measured with - pieces of spectral filters. In such a case, two or more spectral filters must be placed on the solid-state imaging device to cover all 4111 wavelength ranges in total.

However, due to device assembly considerations, a margin for positioning is required between each spectral filter, and unnecessary pixel outputs from the margin portion are mixed in the output of the solid-state image sensor.

By the way, pixel outputs are generally taken into a microcomputer (data processing means) and processed, but in this case, if even unnecessary pixel outputs are taken in,
This amount is redundant, and the RAM as a storage means is wasted.

Regarding these points, the technology shown in the above-mentioned publication,
The number of spectral filters is only one, and the case where multiple spectral filters are required is not considered.

Furthermore, in the previously proposed techniques described above, unnecessary pixels are fixed by a logic circuit, so there is a problem that it is impossible to deal with changes in the positions of unnecessary pixels.

The present invention has been made in view of the above-mentioned problems, and is a combination of a plurality of spectral filters and an integral solid-state image sensor, which captures only effective pixel outputs and eliminates waste in storage means. In addition, it is an object of the present invention to provide a spectroscopic measurement device that does not require the above-mentioned methods and also facilitates the selection and setting of effective pixels and ineffective pixels.

[Means to solve the problem]

The present invention provides a spectroscopy system that includes a plurality of spectral filters that transmit light of different wavelengths depending on the position, a solid-state image sensor that receives light separated by the spectral filter, and a data processing means that controls the solid-state image sensor. In a measurement device, a plurality of spectral filters are arranged discretely on a solid-state image sensor, and the light transmitted between these spectral filters is detected as an invalid pixel of the solid-state image sensor, or as light that passes through the spectral filter. It has an external memory means for storing the position of the effective pixel of the solid-state image pickup device with respect to the incident light, and only the output of the effective pixel is taken into the data processing means.

[Effect]

With this configuration, when pixel outputs are taken into the data processing means, unnecessary pixel data is not taken in, so that only valid pixel data is stored in the RAM in the processing means. Further, even if the positions of valid pixels and invalid pixels change, selection and switching between them can be easily performed using external memory means.

〔Example〕

FIG. 1 shows the configuration of the incident light receiving section in this embodiment.

In the figure, reference numeral 1.2 denotes a spectral filter whose dominant wavelength to be transmitted changes continuously in the X direction of the drawing, and as will be described later, both are arranged discretely for assembly reasons. And, one filter 1 has 390 to 550 (n
m] wavelength band, and the other filter 2 covers a wavelength band of 550 m].
These two filters 1.2 are designed to cover a wavelength band of 390 to 710 (nm).

3 is a CCD image sensor (its schematic configuration is shown in FIG. 1) as an integral solid-state image sensor; filters 1 and 2;
The light 4 incident on the light receiving section is filtered through the filter 1.
, 2, and then enters each pixel of the CCD image sensor 3. Each pixel is irradiated with light of a wavelength corresponding to the main wavelength of the spectral filter 1.2 corresponding to each pixel of the incident light 4, and by reading out the output of each pixel, the incident light 4 is
It is now possible to obtain a spectroscopic spectrum.

Here, the half-value width of the spectral spectrum is determined by the width of change in the dominant wavelength of the spectral filter 1.2 corresponding to the length of each pixel in the X direction in the drawing, and the half-value width at each dominant wavelength of the spectral filters 1 and 2. , and the wavelength pitch of the spectroscopic spectrum is CC
It is determined by the change pitch of the dominant wavelengths of the spectral filters 1 and 2, which corresponds to the pitch of each pixel of the D image sensor 3.

In addition, in this example, each pixel pitch is the dominant wavelength 10 (
nm) pitch.

FIG. 2 shows a specific configuration of the CCD image sensor 3. As shown in FIG. In the figure, 5 is an overflow drain (abbreviated as OD), 6 is an overflow/7'-) (abbreviated as OG), 7 is a photodiode (abbreviated as PD) which is the main part of the CCD image sensor 3, and 8 is a transport Game 1- (SH
9 is a two-phase drive transfer register (abbreviated as RG) that transfers the charge obtained in each pixel.

The PDs 7 are separated by channel stoppers, photoelectrically convert the light incident on each pixel of the CCD image sensor, and further accumulate charges generated by the photoelectric conversion.

Further, SH8 is for taking in the charge generated and accumulated in PD7 to RG9, and when the integration of the image sensor ends 5, by applying a voltage to SH8, PD7
The charge is transferred from RG9 to RG9, and the integration is completed. In addition, when no voltage is applied to OG6, when the incident light is strong, the channel between PD7 is used to prevent the excessive charge accumulated in PD7 from flowing out (overflow) to the adjacent PD7 or RG9. It has a potential lower than that of the stopper and SH8, and discharges excess accumulated charge to OD5 (connected to the power supply voltage and has the lowest potential). Furthermore, when a voltage is applied to the OG6, its potential changes to P.
It is designed to be lower than the potential of D7, and also functions as an integration clear gate to discharge unnecessary charges accumulated in PD7 before integration to OD5 at the start of integration of the CCD image sensor.

10 is a capacitor that converts the charge of each pixel sequentially transferred from RG9 into a voltage, and this capacitor 10 is a reset FET II.
is charged to the power supply voltage prior to charge transfer.

12 is an image sensor output buffer, which outputs the voltage at the capacitor 10 to the outside as a signal O8. 20 is P
This is an aluminum film that blocks light irradiated to D7.

In this embodiment, the pixel located on the rightmost side in FIG. 2 is numbered as pixel 1, and the pixel to the left thereof is numbered as pixel 2. Here, the first pixel shields 15/16 of the area of PD7, and is 1/16 of the normal pixel without shielding PD7.
It is designed to have a light sensitivity of . Similarly, the second pixel has a sensitivity of 1/8 of a normal pixel, the third pixel has a sensitivity of 1/4, the fourth pixel has a sensitivity of 1/2, and the fifth and subsequent pixels are normal pixels with no light shielding. It is called a pixel. Since these light-shielded pixels and the fifth pixel are used as output monitors when switching the integration time, which will be described later, the spectral filter 1.2 is not placed on these pixels. This concludes the description of the configuration of the image sensor.

FIG. 3 shows the block configuration of the control system in this embodiment. The functions of each block and the signal lines between each block will be described below with reference to FIG.

13 is the aforementioned CCD image sensor; 14 is a microcomputer (abbreviated as microcomputer) as a data processing means for controlling the CCD image sensor 13 and processing output data; 15 is an analog processing circuit (details will be described later); 16 is an A/D converter, 17 is a pulse generation circuit,
Reference numeral 18 denotes a basic clock generating section, and this clock generating section 18 generates a basic clock CP1 for generating each pulse signal for driving the image sensor 13 and processing the output signal of the image sensor 13, and an analog output of the image sensor 13. It generates a basic clock CP' for the A/D converter 16 that converts into digital data handled by the microcomputer 14. Further, the pulse generation circuit 17 is composed of logic circuits such as flip-flops, NAND gates, NOR gates, etc., and processes the driving of RG of the image sensor 13, the integral control of the image sensor 13, and the output of the image sensor 13. Each pulse signal for controlling the analog processing circuit 15 is generated.

Each pulse signal will be explained below.

φ1. φ2 is a transfer lock with opposite phases to each other for driving the two-phase drive RG (FIG. 2), and transfer lock φ1. φ
The output of the pixel is taken out in synchronization with one period of φ1-Hi (high), but the charge generated in each pixel is transferred to the capacitor 10 (FIG. 2) at the timing of φ1-Hi (high). φ is a pulse applied to the gate of the reset FET II to charge the above capacitor 5R810 to the power supply voltage prior to charge transfer, and φ
By applying this φ pulse to the above gate at the timing of 1=Lo (low), O3l? S The reset operation of the FET 11 is performed.

φOG and φ8H are OG of the image sensor 13, respectively.

A pulse applied to SH, φOG is a pulse signal for starting integration that becomes Hi before the start of integration and becomes Lo at the time of starting integration, and φ8H transfers the charge accumulated in each PD to RG at the end of integration. When the transfer lock φ1 is Hi, transfer is started as Hi, and when it becomes Lo, it is an integration end signal that ends the integration. These integral time control pulses are generated by the φ signal sent from the microcomputer 14 to the pulse generation circuit 17. My NT controller 14 controls the pulse generation circuit 1 for this integral time control.
The number of transfer locks φ1 transmitted from 7 is counted and constantly monitored.

φI? This is a timing pulse for signal processing in the SS/It O3S/H"7+' processing circuit 1510, φ.

In the output O8 of the CCD image sensor 13, the voltage in the reset state and the voltage in the state where the charge generated in the pixel is transferred appear alternately. It is necessary to extract the signal, apply a gain that can be switched in three stages using the GNI and GN2 signals transmitted from the microcomputer 14 (details will be described later), and perform A/D conversion as Vos. Since the well-known double sampling method (described later) is used to extract the generated charge voltage, φR3S/H'φO3S
/H pulse is supplied to the analog processing circuit 15. Further, φ is an A/D conversion start pulse for A/D converting Vos output from the analog processing circuit 15DS.

The A/D converter 16 starts A/DDS conversion of the output Vos output from the analog processing circuit 15 using the φ pulse sent from the pulse generation circuit 17, and after completing the A/D conversion, performs EOC (End of Conversion). ), and the converted digital data (Data) is transmitted to the microcomputer 14. The microcomputer 14 inputs the output data of each pixel while scanning the EOC, and stores it in the internal RAM.

Reference numerals 101 and 102 designate dip switches (DIP SW) used as external memory means for storing invalid pixel numbers, which will be described later. 201 and 202 are data buses between the microcomputer 14 and each DIP SW, and are connected to input ports 301 and 302 of the microcomputer 14, respectively.

Figure 4 shows the output timing chart of each of the above signals.
The figure shows the operation of the image sensor, and FIG. 6 shows the analog processing circuit. The operation of the image sensor and the signal output timing will be explained with reference to these figures.

At the time of FIG. 4(a), unnecessary charges are accumulated in the PD,
It is in the state shown in FIG. 5(a). After the transfer lock φ1 sent from the pulse generating circuit 17 rises, the microcomputer 14 raises the integration time control signal φ. The pulse generating circuit 17 synchronizes with the rising edge of the NT signal φ and raises the φoG pulse NT (FIG. FF14 (b)). As a result, a voltage is applied to OG, the potential of OG decreases as shown in Figure 5(b), and the unnecessary charge accumulated in PD is
is discharged to. After that, when the pulse of φOG falls (Fig. 4 (C)), as shown in Fig. 5 (C), the OG
The voltage applied to the PD becomes zero, the potential of the OG is restored, and from this point on, the charge generated in the PD begins to be accumulated in the PD, and integration begins. During the integration period, charges are accumulated according to the amount of light incident on each PD (Figure 4 (
d).

Figure 5(d)).

After the microcomputer 14 raises the φ signal, it counts the falling edge of the transfer lock φ1, and after counting the same tarlock φ1 a predetermined number of times, the microcomputer 14 lowers the φ signal after the rising edge of φ1. The pulse NT pulse generating circuit 17 raises the φ8H pulse at the same time as the φ signal falls (FIG. 4(e)).

As a result, as shown in FIG. 5(e), a voltage is applied to SR, the potential of SH decreases, and the charges accumulated in PD are transferred to RG. After that, when the φSH pulse falls (Fig. 4 (r)), Fig. 5 (r)
As shown in , the voltage applied to SH becomes zero, and S
The potential of H is restored, and at this point, the transfer of the charge generated in PD to RG is completed, and the integration of the image sensor is completed. Thereafter, as shown in FIGS. 4(g) and 5(g), the transfer lock φ1. The output charges of each pixel are transferred by φ2 and read out sequentially. After the integration is completed, even if the PD continues to be irradiated with light, the generated charge crosses OG, which has a potential lower than the potential of SH1 or a channel stopper (not shown), and is discharged to OD.

Next, analog processing in the analog processing circuit 15 will be explained. The output O8 of the image sensor 13 is as shown in FIG. The SR3 capacitor is reset to the power supply voltage by the φ pulse generated at the time of the clock φ1mLo, and holds the reset level until the clock φ1 becomes Hi. When the φR8S/It pulse is applied to the reset sample and hold FET 21 shown in FIG. 6, the reset level is held in the sample and hold capacitor 23, and the buffer 2
2 to the non-inverting input of the operational amplifier 24.

When the tarock φ1 becomes Hi, the potential of the output O8 decreases from the reset level by the amount of charge generated in the pixel. This voltage is input to the inverting input of the operational amplifier 24, and output after taking the difference from the reset level previously held by the operational amplifier 24. After this, by the pulse of φ, the above output is passed through the sample O3S/H hold FET'25 to the sample hold capacitor 2.
6 and the operational amplifier 2 via the buffer 27.
It is input to the inverting input of 8. The reason why sample and hold is performed again here is for A/D conversion to be performed later.

The operational amplifier 28 amplifies the image sensor output from the buffer 27 for A/D conversion and outputs it to the A/D converter 16 as Vos. Here, the amplification gain is switched by the GNI and GN2 signals transmitted from the microcomputer 14.

That is, when GN1=GN2-Lo, FET29 is turned on, the gain is -R1/r, and GNI-
In the case of Ht, ON2-Lo, FET30 is turned on,
The gain is -R2/'.

Furthermore, when GNI-GN2=Hi, FET31 is ON.
Therefore, the gain becomes -R3/r.

This concludes the explanation of the operation of the image sensor and the processing of the output of the image sensor.

Next, in this embodiment, a method for obtaining data on the spectroscopic spectrum of incident light will be described.

The two spectral filters 1 and 2 placed on the CCD image sensor 13 are placed in separate areas on the image sensor 13 for easy assembly. 390,400°410...550 of spectral filter 1
The number of each pixel corresponding to the dominant wavelength of each 10 (nm) pitch is N.

N+1. N+2. ......N+16. D.I.P.
The value of N is set in SWI 01.

Before inputting the output of the image sensor 13, the microcomputer 14 reads the value of N set in DIPSWI O1 from the input port 301.

On the other hand, the spectral filter 2 is 560, 570°580...
. . . The number of each pixel corresponding to the dominant wavelength of each 10 (nm) pitch of 710 (nm) is M.

M+1. M+2.・・・・・・M+15. D.I.
The value of M is stored in the PSW 102. Here, the microcomputer 14 reads the value of M set in the DIPSW 102 in the same way as the value of N described above. The two spectral filters 1 and 2 are arranged separately on the image sensor 13, with the pixel (N+16th pixel) corresponding to 550 (nm) of spectral filter 1 and the 560 nm of spectral filter 2.
Pixels that are not used in this system are lined up between the pixel Mth pixel corresponding to [n m)].

In this embodiment, the above-mentioned 1st to 5th pixels are called monitor pixels, each of N pixels to N+16th pixel is called a first group, and the Mth pixel to M+15th pixel is called a second group.
Pixels other than these are called invalid pixels. The process will be explained below according to the flowchart shown in FIG.

First, the microcomputer 14 uses DIPSWI in #100.
The values of N and M are input from 01.102.

Next, in #1, FLG (L) is cleared to O. L takes a value from 1 to 33, where 1 to 17 correspond to pixels N-N+16 in the first group, and 18 to 33 correspond to pixels M-M+15 in the second group. That is, L is 390
This is the number of each output of 10 (nm) pitch up to 710 (nm). In this embodiment, integration is performed by switching the integration time, and from the pixel outputs integrated at each integration time, the pixel output whose output does not saturate and has the highest S/N is selected. At this time, the selected pixel is FLG
(L)-1, all FLGs are
(L) is cleared to 0.

#2 makes it 1-0. 2 is an index of the integration time, and the integration time T is determined in #3. TQ is the maximum integration time during which the output of the image sensor is not saturated. Next, in #4, a shutter (not shown) is opened to irradiate the image sensor with light, and in #5, the above-mentioned
Integrate CD. After the integration is completed, the reading of each pixel output is started, but before this, each counter CN, J, L
Set the value to 1 (#6. #7. #8). The CN value is the pixel number of the image sensor, J is the index for monitor pixels 1 to 5, and L is the wavelength band 390 to 7.
This is the index of the data to be obtained for pixels with a pitch of 10 (nrr+3) up to 10 (nm).

Continuing to step #9, the pixel number CN of the image sensor is determined. Now, if CN≦5, this is determined to be the output of the 1st to 5th monitor pixels, and the process proceeds to #10. #1
At 0, GNI-CN2=L.

As a result, the gain of the operational amplifier 28 of the analog processing circuit 15 is selected to be R1/r, and the outputs of the first to fifth monitor pixels are multiplied by the gain of R1/r. #
These pixel outputs are read in step 11, and data obtained by multiplying this output by 2 is stored in the microcomputer 14 as monitor data MDs (1, J) in step #12. Next, in #13, the monitor pixel counter J is incremented by one, and in #14, the image sensor pixel counter CN is incremented by one.

The routines #9 to #14 are repeated until the CN value reaches 6. When the CN value becomes 6 or more, the result of determination in #9 is No, and the process proceeds to determination in #15.

The CN value is determined in #15, and if the CN value is less than N+16, that is, until the output of the first group is finished, #1 is executed.
Proceed to 7 and GNI-Hi here.

CN2-Lo is output, and the gain of the operational amplifier 28 is R2.
/" is selected. As a result, the output of the first group is multiplied by a gain of R2/".On the other hand, if the CN value becomes larger than N+16, the process proceeds to #16 and GN 1-C
N2-Hi is output, and the gain of the operational amplifier 28 is R3/
r is selected and for the output of the second group R3/r
is multiplied by the gain of As described above, the monitor pixel, the first
The reason why different gains are applied to each of the groups and the second group will be explained below.

The first group has the spectral filter 1 arranged thereon,
The second group has a spectral filter 2 arranged thereon,
No filter is placed on the monitor pixel. The transmittance differs depending on the presence or absence of a filter, and the transmittance for white light differs depending on the spectral filter.

In this way, the level of each output is determined by the monitor, the first group,
Since there are three types in the second group, all output levels are approximately equal for each output, and the gain is switched so as to match the range of the A/D converter. Change the gain above #10° #16. This is done by the microcomputer 14 in #17.

Continuing to explain #18 and subsequent steps. #18 and #19 are determinations for not reading data other than the first group and the second group. That is, as mentioned above, there are unnecessary invalid pixels between the first group and the second group, and between the monitor pixels and the first group, but for these pixels, #18.
No is determined in #19, the process skips steps #20 to #26, and proceeds to #27, where no data is read.

For pixels for which #18 or #19 is YES, that is, pixels in the first group or second group, #2
0, FLG (L) is determined.

Here, if FLG(L)≠0, it means that data No. 5 has already been determined, and #21 to #26
Skip the process and proceed to #27. FLG (L)-
If it is 0, proceed to #21 and read the data.

In #22, the most significant bit (~l5B) of the data is determined. Here, if the MSB is 1, proceed to #24, but
If the MSB is 0, even if the integration time is doubled and the data obtained is doubled, the output will not reach saturation.

Therefore, if the MSB is 00, do not select data here,
When I is increased by one (#55 described below), the integration time is doubled and the M S B is determined again. Also,
In the case of 1-4 in #23, data is selected here in order not to increase I any further. In #24, a flag FLG(L) indicating that this data has been selected is set to 1. In #25, the selected data is multiplied by 2 to normalize the data at each integration time.

That is, as shown in FIGS. 8(a) to (e),
1-0. If you select the data obtained by (T - To), multiply the 12-bit data by 24 to make it 16-bit data as shown in (a), and convert it to -1° (T
If you select the data obtained with -2 It is necessary to shift the data one digit to the right and reduce the data to 1/2. Therefore, multiplying the 12-bit data by 23, b
It16 is set to O to obtain 16-bit data. At this time, data has been obtained up to bit B, and data with an accuracy up to one digit lower than the data obtained in ■-○ can be obtained. The same goes for (C), (d), and (8), and as the integration time increases, up to the lower 16 bits,
Highly accurate data will be obtained. The above processing expands the dynamic range of the image sensor, and enables highly accurate M1 determination even for low output pixels.

Next, in #26, L is increased by one, and the index of the data to be stored is increased. In #27, the pixel number CNO value of the image sensor is increased by one.

In #28, it is determined whether reading of all data has been completed, and if reading of data has not been completed, steps #9 to #
The process of step 27 is repeated, and if the reading is completed, the process moves to the next routine starting from #29.

After #29, the dark output is corrected.

PDs generate charges (dark charges) even when they are not irradiated with light, and the output of each PD is the sum of the light output charges generated in proportion to light irradiation and the dark charges. Spectrometer APj
If an image sensor is used for this purpose, unless this dark charge is corrected, measurement accuracy will deteriorate, especially for low-output pixels. Hereinafter, a specific correction method will be explained along the flowchart.

At #29, a shutter (not shown) is closed to cut off light irradiation. At #30, integration is performed for the integration time determined by #3. Each counter CN, J, L is set to 1 in #31 to #33. In #34, determine CNrli in the same way as above, and if it is less than 5, it is a monitor pixel from 1 to 5, so #
At step 35, GNI-CN2-Lo is output and the gain of the image sensor is set to R1/r. The output of the monitor pixel is read in #36, and the data obtained by multiplying the data by 2 is stored in the memory in the microcomputer 14 as MDD (I, J) in #37. #38. In #39, the J value and CN value are increased by one and the process returns to #34. After the monitor pixel data has been stored, the process moves to step #40.

Here, coefficients are calculated to further calibrate the data obtained at each integration time based on monitor pixel data.

If the amount of incident light changes over time when a total of five integrations are performed by changing the integration time, the output cannot be obtained linearly with respect to the integration time. Therefore, this correction is performed using the output of the monitor pixel.

The first term in the numerator of the above equation is the output of the fifth pixel on the monitor integrated over the integration time To at the time of light irradiation determined in #12, and the second term in the numerator is
The term is the output of the fifth pixel of the monitor, which is integrated over the integration time To in the dark period determined in #37.

By subtracting the second term from the first term, the output of the net generated load corresponding to the light irradiation can be obtained.

The first term in the denominator of the above equation is the output of each pixel of the monitor (5-I) integrated by the integration time ToX2 during light irradiation, obtained in #12, and the second term in the denominator is the output of each pixel in the dark period, obtained in #37. This is the output of each pixel (5-1) of the monitor integrated over an integration time of ToX2. By subtracting the second term from the first term,
Data after dark output correction is obtained. By the above, K
l (0) −1 MDs (2, 3) MDD (2, 3) and coefficients corresponding to each integration time are determined. K1(1) is MD s
MD s (0,
5) MD D(0,5), that is, by dividing the output corresponding to the optical signal integrated by the pixel with sensitivity 1 with the integration time T I NT "" TO, the amount of incident light when the integration time changes The percentage of change is calculated. Kl(2)~
The same applies to Kl(4).

By multiplying the pixel output for each integration time obtained thereafter by 1 (1) by the coefficient for each integration time obtained by the above means, it is possible to eliminate the influence of temporal changes in incident light.

After calculating the coefficient Kl (1) in #40, CN is calculated in #41.
Determine the value, and if CN≦N+16, GN with #42
1=Hi, CN2=Lo is output, and the gain of the image sensor is set to R2/r. If CN>N+16, GNI-CN2-Hi is outputted in #43, and the gain of the image sensor is set to R3/'. #44. If the first and second groups of pixels are output in #45, proceed to #46 and later; if both #44 and #45 are No, then #46 to #5
Skip step 2 and proceed to #53.

Check the FLG (L) at #46, and check this FLG.
If (L) is 1, it means that the pixel was selected in #20 to #25, and the routine proceeds to #47 and later, and FL
If G (L) is not 1, skip #47 to #52 and proceed to #53. In #47, data is read. #
In step 48, the read data is multiplied by 2 and stored as DD(L). This is data for the dark time output correction mentioned above, and is multiplied by the same coefficient 2 in order to match the digit with Ds(L).

In #49, the value obtained by subtracting the DD(L) obtained in #48 from the Dc(L) obtained in #25 is stored as D (L). By #49, the dark output is subtracted from each pixel output, and output data corresponding to the incident light is obtained. In #50, this D (L) is multiplied by the coefficient Kl (1) obtained in #40, and further multiplied by 2 (L). Here 2 (L
) is a coefficient specific to each pixel. Since the sensitivity of each pixel varies, if high-precision measurement is required, it is necessary to make corrections according to this sensitivity. This correction coefficient is stored in advance in the memory within the microcomputer 14. At #50, accurate data for each pixel is stored.

Next, at #51, F4. Put a value of -1 into G (L). FLG (L) is 0, 1. -1, 0 indicates that the data is not finalized, 1 indicates that D
Indicates that the value of s (L) has been determined, and if it is -1, D
Indicates that (L) has been confirmed.

Increment the counter L by 1 in #52, and increment the counter CN in #53.
is incremented by one, and in #54 it is determined whether reading of all pixels of the image sensor has been completed, and if it has not been completed,
Repeat #34 to #53, and if completed, proceed to #55. In #55, increase the integration time index I by one,
If I>4 is not found in #56, repeat #3 to #54 and
> 4, all routines are finished and 390~
The storage of all spectroscopic measurement data of 710 (nm) and 10 (nm) pitches is completed.

This concludes the explanation of the operation of this embodiment.

In this embodiment, the values of the leading pixel numbers N and M of the first and second groups are stored in each DIPSW, which is E2 FROM.

A changeover switch, RAM with backup, and rewritable memory such as FROM are provided, and N is added to this.

The value of M may be stored and taken into the RAM within the microcomputer before the microcomputer reads the data.

Further, in this embodiment, a light irradiation state and a dark state are created by opening and closing a shutter (not shown), but when measuring the reflection spectrum of an object, it is also possible to use the following method. That is, when obtaining the reflection spectrum of an object, a strobe light is used as a light source, a state in which the strobe does not emit light is defined as a dark state, a state in which the strobe emits light is defined as a light irradiation state, and a strobe light is used instead of a mechanical shutter. At this time, the integration time is constant, and the number of flashes of the strobe is 1, 2, 4, 8, and 16.
This is the equivalent of changing the integration time. In this case, the integration time is naturally set to be longer than the time required to emit light 16 times. In this way, the present invention can be similarly applied even when using strobe light.

〔Effect of the invention〕

As described above, according to the present invention, only necessary pixel outputs are taken into the data processing device by the action of the external memory means, so that there is no waste in the RAM as the storage means within the processing device.

In addition, when reading pixel output, data of unnecessary pixels is not taken in, so only valid pixel data is lined up in the RAM in the data processing device. This eliminates the need to take the time and effort to do so, making it possible to save labor and time.

Furthermore, when selecting valid pixels and invalid pixels, the switching can be done using an external memory means, so even if the spectral filter is misaligned during assembly, the value set in the external memory can be changed. This has the effect that it can be easily handled and assembled easily.

[Brief explanation of the drawing]

FIG. 1 is a perspective view showing the configuration of an incident light receiving section of a device according to an embodiment of the present invention, FIG. 2 is a configuration diagram of a CCD image sensor, and FIG. 3 is a block diagram of a control system according to this embodiment. Figure 4 is a timing chart of signal output, Figure 5 is a timing chart of signal output.
Figures (a) to (g) are explanatory diagrams showing the operation of the image sensor, Figure 6 is a circuit diagram of the analog processing circuit, Figure 7 is a control flowchart, and Figures 8 (a) to (e) are data processing FIG. 1.2... Spectral filter, 3, 13... CCD image sensor (solid-state image sensor), 4... Incident light, 14.
...Microcomputer (data processing means), 101
．． 102... Deep switch (external memory means).

Claims (1)

[Scope of Claims] 1. A plurality of spectral filters that transmit light with different wavelengths depending on the position, a solid-state image sensor that receives light separated by the spectral filter, and a data processing means that controls the solid-state image sensor. In a spectroscopic measurement device, a plurality of spectral filters are arranged discretely on a solid-state image sensor, and an ineffective pixel of the solid-state image sensor or a spectral filter for light that passes through between these spectral filters and enters. What is claimed is: 1. A spectroscopic measurement device comprising an external memory means for storing the position of an effective pixel of a solid-state image sensor with respect to light transmitted through and incident on the solid-state image sensor, and only the output of the effective pixel is taken into a data processing means.